SOH and RUL prediction of Li‑ion batteries based on improved Gaussian process regression

Hailin Feng; Guoling Shi

SOH and RUL prediction of Li‑ion batteries based on improved Gaussian process regression

Hailin Feng

Guoling Shi

Vol. 21, No. 12, pp. 1845-1854, Dec. 2021

10.1007/s43236-021-00318-5

Li-ion batteries

State-of-health

Remaining useful life

Gaussian Process Regression

Polynomial regression

Charging current

Abstract

Accurately predicting the state of health (SOH) and remaining useful life (RUL) of Li-ion batteries is the key to Li-ion battery health management. In this paper, a novel GPR-based method for SOH and RUL prediction is proposed. First, five features are extracted from the cyclic charging currents of batteries, and a grey correlation analysis (GRA) shows that these five features are highly correlated with battery capacity. A novel Li-ion battery SOH prediction model is established by improving a basic Gaussian process regression model. Meanwhile, a polynomial regression model is developed to update the feature values in the future. Then the RUL of a battery is predicted by combining the SOH prediction model. Finally, the prediction effect of the proposed model is compared with other models using four Li-ion battery degradation data. The obtained results show that the model proposed in this paper has the highest accuracy. The robustness of the proposed model is verified by random walk battery data.

Statistics

Show / Hide Statistics

Cumulative Counts from September 30th, 2019
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.

Cite this article

[IEEE Style]

H. Feng and G. Shi, "SOH and RUL prediction of Li‑ion batteries based on improved Gaussian process regression," Journal of Power Electronics, vol. 21, no. 12, pp. 1845-1854, 2021. DOI: 10.1007/s43236-021-00318-5.

[ACM Style]

Hailin Feng and Guoling Shi. 2021. SOH and RUL prediction of Li‑ion batteries based on improved Gaussian process regression. Journal of Power Electronics, 21, 12, (2021), 1845-1854. DOI: 10.1007/s43236-021-00318-5.